Atomist or Holist? A Diagnosis and Vision for More Productive Interdisciplinary AI Ethics Dialogue

Travis Greene, Amit Dhurandhar, Galit Shmueli

Introduction

The field of Artificial Intelligence (AI) is currently grappling with a profound internal schism. As major research venues like NeurIPS began requiring "ethics impact statements," the community fractured. While some heralded these requirements as a necessary step toward social responsibility, others decried them as "repressive actions" that threaten academic freedom and scientific progress. This paper argues that these heated debates are not merely administrative disagreements but symptoms of a fundamental clash between two distinct worldviews: Atomism and Holism.

Methods: A Conceptual Diagnosis

This research employs a conceptual and philosophical framework to diagnose the current state of AI ethics discourse. Rather than an empirical study, it utilizes an Ideological Taxonomy to categorize researchers based on their underlying beliefs. By applying Thomas Kuhn’s concept of "scientific paradigms," we explain why many technical researchers view themselves as "puzzle solvers" operating within a stable, value-neutral framework.

Furthermore, we draw on C.P. Snow’s 1959 "Two Cultures" thesis, which identified a growing divide between the sciences and the humanities. In the modern AI era, this divide has manifested as a conflict between those who prioritize technical optimization (Atomists) and those who prioritize social impact (Holists). To resolve this, we adapt Carl Rogers’ "person-centered approach" to psychotherapy, creating a model for interdisciplinary dialogue centered on mutual understanding.

The Ideological Divide: Atomism vs. Holism

To diagnose this polarization, we propose a taxonomy rooted in philosophy and intellectual history. At the heart of the conflict lies the "is-ought" problem—the question of whether scientific facts can be separated from human values.

Atomists operate within a paradigm where science is viewed as value-neutral "puzzle-solving." In this view, the researcher's role is to solve technical problems within a defined framework, leaving the ethical implications to policymakers or society at large. Conversely, Holists view science as inextricably tied to social values. They see themselves as "social stewards" or fiduciaries, obligated to communicate and mitigate the risks inherent in their privileged technical knowledge.

Comparison of Atomist and Holist worldviews across philosophical and social dimensions.

As shown in the table above, these worldviews diverge on nearly every fundamental level, from their guiding metaphors to their scientific methodologies. Atomists prioritize individual rights and impartial justice, fearing that subjective ethics standards will lead to dogmatism and "wishful thinking" that kills innovation. Holists prioritize empathy and collective well-being, fearing that "value-neutral" science creates a faceless, unaccountable technocracy focused solely on prediction and control.

A Vision for Productive Dialogue

Moving past name-calling and "cancellation" threats requires a structured approach to interdisciplinary dialogue. We propose a model based on Rogerian psychotherapy, emphasizing "Accuracy and Empathy." In this framework, participants must accurately restate their opponent's position to the opponent's satisfaction before offering a rebuttal. This ensures that the dialogue is grounded in mutual respect rather than straw-man arguments.

Beyond dialogue, we identify four targeted strategies to bridge the gap between technical researchers and humanities scholars, tailored to their specific disciplinary backgrounds:

  1. Moral Education for Data Scientists: Helping technical researchers understand the historical and philosophical context of the fact-value distinction.
  2. Incentives for Economists: Using mechanisms like prizes for creative ethics statements to encourage deeper engagement with social impact.
  3. Philosophical and Legal Training for Technology Activists: Moving beyond "mob-based" critiques toward rigorous, framework-based arguments.
  4. Technical Training for Humanities Scholars: Ensuring that ethical critiques are grounded in the technical realities of AI development, perhaps through cross-disciplinary internships.
A conceptual bridge between Atomists and Holists, mediated by Accuracy, Empathy, and targeted educational strategies.

Conclusion and Limitations

While this framework provides a path forward, it faces significant challenges. The "therapeutic" dialogue model may struggle in the high-heat environment of social media (Twitter/X), where nuance is often lost. Furthermore, while the Atomist-Holist binary is a powerful diagnostic tool, it is an oversimplification of a more complex spectrum of beliefs held by researchers.

Future work must address the role of corporate power in shaping these dialogues and develop formal accreditation processes for ethics reviewers to ensure they possess the necessary interdisciplinary expertise. By fostering a shared vocabulary and mutual empathy, the AI community can move toward a more productive synthesis that ensures technological progress truly benefits the public.

Summary updated

BibTeX
@article{greene2023atomist,
  author = {Travis Greene and Amit Dhurandhar and Galit Shmueli},
  title = {{Atomist or Holist? A Diagnosis and Vision for More Productive Interdisciplinary AI Ethics Dialogue}},
  journal = {Patterns},
  year = {2023},
  doi = {10.1016/j.patter.2022.100652},
}
Figure